常规稻与杂交稻谷的仿生电子鼻分类识别

气味是进行稻谷品种及其品质识别的重要方法之一,作为一种基于仿生嗅觉的机器检测方法,仿生电子鼻在水稻品种的分类识别中具有较好的应用前景。常规稻与杂交稻在食味品质等方面存在一定的差异,为了解应用电子鼻进行常规稻谷与杂交稻谷识别的可行性,采用PEN3电子鼻对同季同地域收获的3种常规稻(中香1号、湘晚13、瑶平香)和3种杂交稻(伍丰优T025、品36、优优122)稻谷样品的气味信息进行了采集和分析。首先通过过载分析(Loadings)法分析了电子鼻检测稻谷气体挥发物时的各传感器贡献率,分别针对基于特征值的提取和稻谷气味检测对电子鼻传感器阵列中的传感器进行了优选,阐明了稻谷气体挥发物检测中应以对硫化物、氮氧化合物、芳香成分和有机硫化物敏感的传感器为主。随后,分别采用主成分分析法(principal component analysis,PCA)、线性判别法(linear discriminant analysis,LDA)和BP神经网络对6种不同稻谷之间、常规稻与杂交稻之间的分类识别进行了研究。结果表明,PCA分析法与LDA分析法在对6种不同稻谷之间的分类以及常规稻与杂交稻之间的分类中均未取得理想的效果,存在部分样本数据点重叠或样本数据点较近的情况,在实际应用中易发生混淆;而BP神经网络在对6种不同稻谷之间的分类中对测试集的识别正确率分别达到了90%,在常规稻与杂交稻之间的分类识别中对测试集的识别正确率达到了96.7%。上述试验验证了电子鼻用于常规稻与杂交稻稻谷分类识别的有效性,为常规稻与杂交稻的快速、无损分类识别提供了一种新的方法。

Classification and recognition of hybrid and inbred rough rice based on bionic electronic nose

Abstract: The bionic electronic nose, a machine detection method based on bionic olfaction, enjoys a good application prospect in rice varieties classification and recognition. There are many differences between hybrid and inbred rice. In order to understand the feasibility by using an electronic nose to classify and recognize hybrid and inbred rough rice varieties, the samples' volatiles of 3 inbred rough rice varieties (Zhongxiang 1, Xiangwan 13, Yaopingxiang) and 3 hybrid rough rice varieties (Wufengyou T025, Pin 36, Youyou 122), which grow in the same area and the same season, were collected in this article by using an electronic nose (PEN3). Firstly, the contribution rates of sensors, which in sampling rough rice volatile, were analyzed by using loadings analysis, and the electronic nose's sensors in the array were selected based on feature value extraction and rough rice volatile detection. It is indicated that the sensors are keenly sensitive to sulfur-containing organics, nitrogen oxides, aromatics, and sulfur- and chlorine-containing organics, and it should be mainly used in classified rough rice varieties. After that, classification and recognition algorithms of Hybrid and Inbred Rough Rice, including PCA (principal component analysis), LDA (linear discriminant analysis), and BP neural network analysis, were developed. Results show that PCA and LDA analysis for the classification between 6 rough rice varieties did not achieve the ideal results. Neither did the classification between hybrid and inbred rough rice. There are some overlapping regions between the classification groups. It is easy to cause a blur in practical application. Compared with PCA and LDA, BP neural network has better performance in the classification of 6 different rough rice varieties, the same effect in the classification between hybrid and inbred rough rice. By using BP neural network, test results show that the accuracy of classification between 6 different rough rice varieties reaches 90% in testing samples test. For classification between hybrid and inbred rough rice, it reaches 96.7% in testing samples test. It is indicated that the bionic Electronic Nose could be a new approach, which can conduct a rapid and non-destructive classification of hybrid and inbred rough rice varieties.